Introduction: The ability to discriminate among ligands binding to the same protein target in terms of their relative binding affinity lies at the heart of structure-based drug design. Any improvement in the accuracy and reliability of binding affinity prediction methods decreases the discrepancies between experimental and computational results. Objectives: The primary objectives were to find the most relevant features affecting binding affinity, least use of manual feature engineering, and improving the reliability of binding affinity prediction using deep learning models through the systematic comparison of representative deep learning architectures and improving the model performance by tuning the hyperparameters. Methods: The binding site of target proteins was represented as a grid box around their bound ligand. Both binary and distance-dependent occupancies were applied for how an atom affects its neighbor voxels in this grid. A combination of different features including ANOLEA, ligand elements, and Arpeggio atom types were used to represent the input. Several convolutional neural networks (CNNs) with different architectures were trained and tested on the 2016 PDBbind dataset, including VGG, ResNet, SqueezeNet, MobileNet, and ShuffleNet. Four independent challenging datasets were additionally used to evaluate the models. Results: The best tuned model showed an improved accuracy in the binding affinity prediction on PDBbind core subset (Pearson’s R=0.83) and is better than the recent state-of-the-art models in this field. In addition when the model was trained on our proposed benchmark dataset, it reaches Pearson correlation coefficient of 0.86 which confirms the value of our extended dataset. Conclusions: The promising results for the predicted binding affinities is expected to pave the way for embedding deep learning models in virtual screening and rational drug design fields.

Improving the prediction of protein-ligand binding affinity using deep learning models

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.